Fire-prone dry forests often face increasing fires from climate change with low resistance and resilience due to logging of large, old fire-resistant trees. Their restoration across large landscapes is constrained by limited mature trees, physical settings, and protection. Active restoration has been costly and shown limited effectiveness, but lower cost passive restoration is less studied. I used GIS and machine learning to see if passive restoration of old trees could overcome constraints in time, by 2060, across 667,000 ha of montane forests in the San Juan Mountains, Colorado, where temperatures are increasing faster than the global average. Random Forest models of physical locations of reconstructed historical old growth (OG) and relatively frequent fire (RFF) show historical OG with RFF was favored between 6.1-7.9̊C annual mean temperatures. Random Forest models projected similar temperature-suitable locations were moved into the current middle montane ca 2015, and would be extended to just below the upper limit of the montane if the Paris 1.5̊C goal is reached, but beyond if not. US Forest Service Common Stand Exam data, which covered ∼15% of the study area and included 26,149 tree ages, show the highest potential for restoring resistance and resilience from old trees is a ≥120 year age class. This class could become a ≥160 year age class, which meets old-growth age criteria, over 81% of the area by ca 2060, nearly fully restoring historical old-growth levels. Half this age class is already protected, and much of the remainder could be retained using evidence-based diameter caps. Datasets thus are sufficient to show that passive restoration of old-tree resistance and resilience to fire is feasible by ca 2060 across a large montane landscape, although contingent on global success in achieving the Paris 1.5̊C goal. Passive restoration may be viable elsewhere.

Derived using GIS and Random Forest models from historical and modern datasets.

See pdf that explains the datasets.